Expanderizing Higher Order Random Walks

Vedat Levi Alev,Shravas Rao
2024-06-03
Abstract:We study a variant of the down-up and up-down walks over an $n$-partite simplicial complex, which we call expanderized higher order random walks -- where the sequence of updated coordinates correspond to the sequence of vertices visited by a random walk over an auxiliary expander graph $H$. When $H$ is the clique, this random walk reduces to the usual down-up walk and when $H$ is the directed cycle, this random walk reduces to the well-known systematic scan Glauber dynamics. We show that whenever the usual higher order random walks satisfy a log-Sobolev inequality or a Poincaré inequality, the expanderized walks satisfy the same inequalities with a loss of quality related to the two-sided expansion of the auxillary graph $H$. Our construction can be thought as a higher order random walk generalization of the derandomized squaring algorithm of Rozenman and Vadhan. We show that when initiated with an expander graph our expanderized random walks have mixing time $O(n \log n)$ for sampling a uniformly random list colorings of a graph $G$ of maximum degree $\Delta = O(1)$ where each vertex has at least $(11/6 - \epsilon) \Delta$ and at most $O(\Delta)$ colors and $O\left( \frac{n \log n}{(1 - \| J\|)^2}\right)$ for sampling the Ising model with a PSD interaction matrix $J \in R^{n \times n}$ satisfying $\| J \| \le 1$ and the external field $h \in R^n$-- here the $O(\bullet)$ notation hides a constant that depends linearly on the largest entry of $h$. As expander graphs can be very sparse, this decreases the amount of randomness required to simulate the down-up walks by a logarithmic factor. We also prove some simple results which enable us to argue about log-Sobolev constants of higher order random walks and provide a simple and self-contained analysis of local-to-global $\Phi$-entropy contraction in simplicial complexes -- giving simpler proofs for many pre-existing results.
Data Structures and Algorithms
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is to study and analyze the mixing time of expanderized higher - order random walks under specific conditions and their relationship with standard higher - order random walks by introducing expanderized higher - order random walks. Specifically, the paper focuses on the following aspects: 1. **Defining expanderized higher - order random walks**: The paper defines a new random walk method - expanderized down - up and up - down walks. These walks are carried out on an auxiliary expander graph \(H\). When \(H\) is a clique, this walk degenerates into the traditional down - up walk; when \(H\) is a directed cycle, this walk degenerates into a systematic scan. 2. **Analyzing the properties of expanderized walks**: The paper proves that if the traditional higher - order random walk satisfies the log - Sobolev inequality or the Poincaré inequality, then the expanderized walk also satisfies the same inequality, but with a decrease in quality, and the degree of decrease is related to the two - way expandericity of the auxiliary graph \(H\). 3. **Studying the mixing time of expanderized walks**: The paper studies the mixing time of expanderized walks in two specific examples: - For a graph \(G\) with maximum degree \(\Delta = O(1)\), in the case where each vertex has at least \((11/6-\epsilon)\Delta\) colors and at most \(O(\Delta)\) colors, the mixing time of the expanderized down - up walk is \(O(n\log n)\). - For the Ising model with a positive semi - definite interaction matrix \(J\) and an external field \(h\), under the condition that \(\|J\|_{\text{op}}\leq1\), the mixing time of the expanderized down - up walk is \(O\left(\frac{n\log n}{(1 - \|J\|_{\text{op}})^2}\right)\). 4. **Providing simplified proof tools**: The paper also provides some simple results, enabling the author to argue the log - Sobolev constant of the higher - order random walk, and gives a simple self - contained analysis of local - to - global \(\Phi\)-entropy contraction, simplifying the proofs of many existing results. ### Main contributions - **Expanderized higher - order random walks**: A new random walk method is introduced, which combines the characteristics of expander graphs and can maintain good mixing performance while reducing the need for randomness. - **Theoretical analysis**: Through strict mathematical proofs, the relationship between the mixing time of expanderized walks and traditional higher - order random walks is shown. - **Application examples**: The effectiveness of expanderized walks is verified on two specific problems, graph coloring and the Ising model, showing its potential in practical applications. ### Formula presentation - **Spectral Gap**: \[ \text{gap}(P)=1 - \lambda(P) \] where \(\lambda(P)=\max\{\lambda_2(P),|\lambda_{\min}(P)|\}\). - **Mixing Time**: \[ \tau_{\text{mix}}(P,\epsilon)\leq\frac{1}{1 - \lambda(P)}\cdot\log\left(\frac{1}{\epsilon\sqrt{\min_{\omega\in\Omega}\pi(\omega)}}\right) \] - **Entropy Contraction Constant**: